In trying to create a new file format, several sets of conflicting and mutually contradictory goals must somehow be reconciled.
For example, the files should be completely self-documenting, but that entails writing all sorts of information which will cause the file size to swell. On the other hand, the data should be compact, since there will be many files, and it is desirable to conserve disk space.
One would like the data to be portable, so that files can be moved from machine to machine without having to convert the data. However, reading the files ought to be efficient, to minimize run time and CPU charges.
Ideally, the new data files would adhere to some sort of universal standard. Individual projects and users, though, should be able to customize aspects of the files to meet their special needs.
Any file format specification reflects a tradeoff or compromise between these (and possibly other) sets of mutually contradictory goals. We wish to find the file format whose compromises best meet our needs.